Reactivation of stalled replication forks requires specialized mechanisms that can recognize the fork structure and promote downstream processing events. Fork regression has been implicated in several models of fork reactivation as a crucial processing step that supports repair. However, it has also been suggested that regressed forks represent pathological structures rather than physiological intermediates of repair. To investigate the biological role of fork regression in bacteriophage T4, we tested several mechanistic models of regression: strand exchange-mediated extrusion, topology-driven fork reversal and helicase-mediated extrusion. Here, we report that UvsW, a T4 branch-specific helicase, is necessary for the accumulation of regressed forks in vivo, and that UvsW-catalysed regression is the dominant mechanism of origin-fork processing that contributes to double-strand end formation. We also show that UvsW resolves purified fork intermediates in vitro by fork regression. Regression is therefore part of an active, UvsW-driven pathway of fork processing in bacteriophage T4.